Thanks, Tejinder. I am honored that you ask.
The gist, following Wheeler ("it from bit") is that there is zero distance between a yes-no question and its answer, no matter how separated in time; in other words, a relativistic "finite and unbounded" reality does not have to be finite in time and unbounded in space -- general relativity suffers no loss of generality in a model finite in space and unbounded in time. The binary relation holds in either case.
This can be made rigorous:
For reasons explained in the essay, we assign positive-definite values 1/2 to yes, 1/4 to no; indefinite values of - 1/2 to not-yes and - 1/4 to not-no.
In a Bell-Aspect type experiment, 1/2 1/4 = 3/4 describes the upper limit of probability (75%) that Alice and Bob will have correlated answers, when they decide on a cooperative strategy in advance. Random correlations will fall to the average of correlations, 1/2, by the pair of equations:
1/2 1/4 - 1/2 1/4 = 1/2
- 1/2 - 1/4 1/2 - 1/4 = - 1/2
Because there is no negative probability, the prior assumption of a probability measure function adds a priori an extra sign, such that:
( 1/2) 1/4 - 1/2 1/4 = 1/2
- (- 1/2) - 1/4 1/2 - 1/4 = 1/2
So -- as it should be -- assuming that nature is fundamentally probabilistic, the singular average 1/2 applies in both cases. Underlying this calculation is the implied assumption of probability theory: the hypothesis of equally likely outcomes. That's why the extra sign, which brings with it the additional implication of a non-orientable space.
The contest asks us to identify and question foundational assumptions -- one of the most persistent of these is probabilistic measure. When we eliminate that assumption, what's left is topological orientability, left hand and right hand. That's the reason that I say the distance between question and answer is zero -- because orientability implies an additional degree of freedom by which no matter if the initial answer is yes followed by no (3/4)or not-yes followed by not-not-no, the outcomes is zero. (Thus, zero distance between question and answer, a result that can only come of topological analysis, where distance carries a different meaning than in ordinary geometry.) Thus:
1/2 1/4 - 1/2 1/4 - 1/2 = 0 (Left Hand, positive rotation in the plane)
- 1/2 - 1/4 1/2 - 1/4 1/2 = 0 (Right Hand, negative rotation in the plane)
The sign pair, and - -, are the same initial condition as the probabilistic model, except that the initial condition is compelled to be orientable, i.e., yes no (or not-yes minus not-not-no which is sign-reverse equivalent) such that an equal number of measurement events in the orientable space as the probability space, gives a zero remote outcome, implying Right Hand and Left Hand variables.
So instead of a probability average 1/2 for quantum correlated events, dependent on observer orientation and implying an observer-created probabilistic reality, we get a classical continuous wave function, with no probability function at all. The range of continuous values of the wave function are dichotomous correlated discrete values -- binary -- just as Wheeler said.
Therefore:
Local, physical information of a remote measurement outcome is compelled to originate from a point at infinity, because it's that singularity which distinguishes the nonorientable measure space of R^3 from the topology of S^3, the orientable space that is the source of continuous binary variables.
Hope this helps. I was just going to post the "gist," but I couldn't stop myself, because I think the argument is quite elegant.
Best,
Tom